-; RUN: llc -O3 -disable-peephole -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+avx,+f16c < %s | FileCheck %s
+; RUN: llc -O3 -disable-peephole -mtriple=x86_64-unknown-unknown -mattr=+avx,+f16c < %s | FileCheck %s
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-unknown"
%4 = xor <2 x i64> %2, <i64 -1, i64 -1>
%5 = and <2 x i64> %4, %3
%6 = bitcast <2 x i64> %5 to <4 x float>
- ret <4 x float> %6
+ ; fadd forces execution domain
+ %7 = fadd <4 x float> %6, <float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <4 x float> %7
}
define <8 x float> @stack_fold_andnps_ymm(<8 x float> %a0, <8 x float> %a1) {
%4 = xor <4 x i64> %2, <i64 -1, i64 -1, i64 -1, i64 -1>
%5 = and <4 x i64> %4, %3
%6 = bitcast <4 x i64> %5 to <8 x float>
- ret <8 x float> %6
+ ; fadd forces execution domain
+ %7 = fadd <8 x float> %6, <float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <8 x float> %7
}
define <2 x double> @stack_fold_andpd(<2 x double> %a0, <2 x double> %a1) {
%3 = bitcast <4 x float> %a1 to <2 x i64>
%4 = and <2 x i64> %2, %3
%5 = bitcast <2 x i64> %4 to <4 x float>
- ret <4 x float> %5
+ ; fadd forces execution domain
+ %6 = fadd <4 x float> %5, <float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <4 x float> %6
}
define <8 x float> @stack_fold_andps_ymm(<8 x float> %a0, <8 x float> %a1) {
%3 = bitcast <8 x float> %a1 to <4 x i64>
%4 = and <4 x i64> %2, %3
%5 = bitcast <4 x i64> %4 to <8 x float>
- ret <8 x float> %5
+ ; fadd forces execution domain
+ %6 = fadd <8 x float> %5, <float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <8 x float> %6
}
define <2 x double> @stack_fold_blendpd(<2 x double> %a0, <2 x double> %a1) {
define <4 x float> @stack_fold_extractf128(<8 x float> %a0, <8 x float> %a1) {
;CHECK-LABEL: stack_fold_extractf128
;CHECK: vextractf128 $1, {{%ymm[0-9][0-9]*}}, {{-?[0-9]*}}(%rsp) {{.*#+}} 16-byte Folded Spill
- ;CHECK: vmovaps {{-?[0-9]*}}(%rsp), %xmm0 {{.*#+}} 16-byte Reload
%1 = shufflevector <8 x float> %a0, <8 x float> %a1, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
%2 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
ret <4 x float> %1
ret <8 x float> %2
}
-; TODO stack_fold_insertps
+define <4 x float> @stack_fold_insertps(<4 x float> %a0, <4 x float> %a1) {
+ ;CHECK-LABEL: stack_fold_insertps
+ ;CHECK: vinsertps $17, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
+ ;CHECK-NEXT: {{.*#+}} xmm0 = zero,mem[0],xmm0[2,3]
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
+ %2 = call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %a0, <4 x float> %a1, i8 209)
+ ret <4 x float> %2
+}
+declare <4 x float> @llvm.x86.sse41.insertps(<4 x float>, <4 x float>, i8) nounwind readnone
define <2 x double> @stack_fold_maxpd(<2 x double> %a0, <2 x double> %a1) {
;CHECK-LABEL: stack_fold_maxpd
%3 = bitcast <4 x float> %a1 to <2 x i64>
%4 = or <2 x i64> %2, %3
%5 = bitcast <2 x i64> %4 to <4 x float>
- ret <4 x float> %5
+ ; fadd forces execution domain
+ %6 = fadd <4 x float> %5, <float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <4 x float> %6
}
define <8 x float> @stack_fold_orps_ymm(<8 x float> %a0, <8 x float> %a1) {
%3 = bitcast <8 x float> %a1 to <4 x i64>
%4 = or <4 x i64> %2, %3
%5 = bitcast <4 x i64> %4 to <8 x float>
- ret <8 x float> %5
+ ; fadd forces execution domain
+ %6 = fadd <8 x float> %5, <float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <8 x float> %6
}
define <8 x float> @stack_fold_perm2f128(<8 x float> %a0, <8 x float> %a1) {
}
declare <8 x float> @llvm.x86.avx.round.ps.256(<8 x float>, i32) nounwind readnone
-; TODO stack_fold_roundsd
+define double @stack_fold_roundsd(double %a0) optsize {
+ ;CHECK-LABEL: stack_fold_roundsd
+ ;CHECK: vroundsd $9, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 8-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
+ %2 = call double @llvm.floor.f64(double %a0)
+ ret double %2
+}
+declare double @llvm.floor.f64(double) nounwind readnone
; TODO stack_fold_roundsd_int
declare <2 x double> @llvm.x86.sse41.round.sd(<2 x double>, <2 x double>, i32) nounwind readnone
-; TODO stack_fold_roundss
+define float @stack_fold_roundss(float %a0) optsize {
+ ;CHECK-LABEL: stack_fold_roundss
+ ;CHECK: vroundss $9, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 4-byte Folded Reload
+ %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
+ %2 = call float @llvm.floor.f32(float %a0)
+ ret float %2
+}
+declare float @llvm.floor.f32(float) nounwind readnone
; TODO stack_fold_roundss_int
declare <4 x float> @llvm.x86.sse41.round.ss(<4 x float>, <4 x float>, i32) nounwind readnone
define <4 x float> @stack_fold_shufps(<4 x float> %a0, <4 x float> %a1) {
;CHECK-LABEL: stack_fold_shufps
- ;CHECK: vshufps $-56, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
+ ;CHECK: vshufps $200, {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
%2 = shufflevector <4 x float> %a0, <4 x float> %a1, <4 x i32> <i32 0, i32 2, i32 4, i32 7>
ret <4 x float> %2
define <8 x float> @stack_fold_shufps_ymm(<8 x float> %a0, <8 x float> %a1) {
;CHECK-LABEL: stack_fold_shufps_ymm
- ;CHECK: vshufps $-108, {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
+ ;CHECK: vshufps $148, {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
- %2 = shufflevector <8 x float> %a0, <8 x float> %a1, <8 x i32> <i32 0, i32 1, i32 9, i32 10, i32 undef, i32 undef, i32 undef, i32 undef>
+ %2 = shufflevector <8 x float> %a0, <8 x float> %a1, <8 x i32> <i32 0, i32 1, i32 9, i32 10, i32 4, i32 5, i32 13, i32 14>
ret <8 x float> %2
}
;CHECK: vunpckhpd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
%2 = shufflevector <2 x double> %a0, <2 x double> %a1, <2 x i32> <i32 1, i32 3>
- ret <2 x double> %2
+ ; fadd forces execution domain
+ %3 = fadd <2 x double> %2, <double 0x0, double 0x0>
+ ret <2 x double> %3
}
define <4 x double> @stack_fold_unpckhpd_ymm(<4 x double> %a0, <4 x double> %a1) {
;CHECK: vunpckhpd {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
%2 = shufflevector <4 x double> %a0, <4 x double> %a1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
- ret <4 x double> %2
+ ; fadd forces execution domain
+ %3 = fadd <4 x double> %2, <double 0x0, double 0x0, double 0x0, double 0x0>
+ ret <4 x double> %3
}
define <4 x float> @stack_fold_unpckhps(<4 x float> %a0, <4 x float> %a1) {
;CHECK: vunpckhps {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
%2 = shufflevector <4 x float> %a0, <4 x float> %a1, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
- ret <4 x float> %2
+ ; fadd forces execution domain
+ %3 = fadd <4 x float> %2, <float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <4 x float> %3
}
define <8 x float> @stack_fold_unpckhps_ymm(<8 x float> %a0, <8 x float> %a1) {
;CHECK: vunpckhps {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
%2 = shufflevector <8 x float> %a0, <8 x float> %a1, <8 x i32> <i32 2, i32 10, i32 3, i32 11, i32 6, i32 14, i32 7, i32 15>
- ret <8 x float> %2
+ ; fadd forces execution domain
+ %3 = fadd <8 x float> %2, <float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <8 x float> %3
}
define <2 x double> @stack_fold_unpcklpd(<2 x double> %a0, <2 x double> %a1) {
;CHECK: vunpcklpd {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
%2 = shufflevector <2 x double> %a0, <2 x double> %a1, <2 x i32> <i32 0, i32 2>
- ret <2 x double> %2
+ ; fadd forces execution domain
+ %3 = fadd <2 x double> %2, <double 0x0, double 0x0>
+ ret <2 x double> %3
}
define <4 x double> @stack_fold_unpcklpd_ymm(<4 x double> %a0, <4 x double> %a1) {
;CHECK: vunpcklpd {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
%2 = shufflevector <4 x double> %a0, <4 x double> %a1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
- ret <4 x double> %2
+ ; fadd forces execution domain
+ %3 = fadd <4 x double> %2, <double 0x0, double 0x0, double 0x0, double 0x0>
+ ret <4 x double> %3
}
define <4 x float> @stack_fold_unpcklps(<4 x float> %a0, <4 x float> %a1) {
;CHECK: vunpcklps {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
%2 = shufflevector <4 x float> %a0, <4 x float> %a1, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
- ret <4 x float> %2
+ ; fadd forces execution domain
+ %3 = fadd <4 x float> %2, <float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <4 x float> %3
}
define <8 x float> @stack_fold_unpcklps_ymm(<8 x float> %a0, <8 x float> %a1) {
;CHECK: vunpcklps {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
%2 = shufflevector <8 x float> %a0, <8 x float> %a1, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 4, i32 12, i32 5, i32 13>
- ret <8 x float> %2
+ ; fadd forces execution domain
+ %3 = fadd <8 x float> %2, <float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <8 x float> %3
}
define <2 x double> @stack_fold_xorpd(<2 x double> %a0, <2 x double> %a1) {
%3 = bitcast <4 x float> %a1 to <2 x i64>
%4 = xor <2 x i64> %2, %3
%5 = bitcast <2 x i64> %4 to <4 x float>
- ret <4 x float> %5
+ ; fadd forces execution domain
+ %6 = fadd <4 x float> %5, <float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <4 x float> %6
}
define <8 x float> @stack_fold_xorps_ymm(<8 x float> %a0, <8 x float> %a1) {
%3 = bitcast <8 x float> %a1 to <4 x i64>
%4 = xor <4 x i64> %2, %3
%5 = bitcast <4 x i64> %4 to <8 x float>
- ret <8 x float> %5
+ ; fadd forces execution domain
+ %6 = fadd <8 x float> %5, <float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0, float 0x0>
+ ret <8 x float> %6
}